New AI model revolutionises disease diagnosis with visual maps

Scientists have developed an artificial intelligence (AI) model, which has the remarkable ability to accurately identify tumors and diseases in medical images. This innovative model provides a unique level of transparency by explaining each diagnosis with a visual map, allowing doctors to easily follow its line of reasoning, double-check for accuracy, and explain the results to patients.

Sourya Sengupta, a graduate student at Beckman Institute for Advanced Science and Technology in the US and the lead author of the study, emphasized the significance of this development, stating, “The idea is to help catch cancer and disease in its earliest stages — like an X on a map — and understand how the decision was made. Our model will help streamline that process and make it easier on doctors and patients alike,” said Sengupta.

The new AI model’s transparency is expected to have a significant impact on the process of decoding medical images, particularly in regions with a scarcity of doctors and long patient queues. Sengupta highlighted the potential of AI in such scenarios, stating, “When time and talent are in high demand, automated medical image screening can be deployed as an assistive tool — in no way replacing the skill and expertise of doctors.”

The model’s ability to pre-scan medical images and flag those containing something unusual, such as a tumor or early sign of disease, for a doctor’s review is expected to save time and improve the performance of the person tasked with reading the scan.

The researchers trained their model on three different disease diagnosis tasks, including more than 20,000 images. The model demonstrated impressive performance, with accuracy rates of 77.8% for mammograms, 99.1% for retinal optical coherence tomography (OCT) images, and 83% for chest X-rays. These high accuracy rates are attributed to the AI’s deep neural network, which mimics the nuance of human neurons in making decisions.

The model’s performance was compared to existing AI systems, and it performed comparably in all three categories, showcasing its potential to revolutionize disease diagnosis through medical imaging.

The development of this new AI model represents a significant advancement in the field of medical imaging and disease diagnosis. Its unique transparency and accuracy have the potential to streamline the process of identifying tumors and diseases in medical images, ultimately benefiting both doctors and patients.

This innovative approach to disease diagnosis is expected to have a profound impact on the early detection and understanding of various medical conditions, marking a significant milestone in the intersection of AI and healthcare.